The model adds $4,810 for each additional person to Hawaii's 2018 poverty guideline, where HHS adds $4,970 #10894

For 2018, the model builds Hawaii's poverty guideline from a $13,960 first-person amount plus $4,810 for each additional person, the amount HHS used in 2017. HHS adds $4,970 for each person after the first in its 2018 Hawaii table, so the model's guideline falls short by $160 for two people, $480 for four and $1,120 for eight.

We also list this gap in #10888, in its table of other years.

What HHS publishes

HHS published the 2018 guidelines in the Federal Register on January 18, 2018 (83 FR 2643). For Hawaii, HHS sets the guideline at $13,960 for one person and $48,750 for eight, and adds $4,970 for each person after the first. The IRS prints the same table as Table 1-3 of its 2019 Form 8962 instructions and directs filers to use the 2018 guidelines for the 2019 premium tax credit.

Who it reaches

Current-year results don't change. In any 2018 calculation, the model reads these values for every Hawaii unit of two or more. It also reads them in programs that lag the guideline:

• the premium tax credit for 2019
• WIC from July 2018 through June 2019
• SNAP income standards and tanf_fpg from October 2018 through September 2019

For example, take a Hawaii couple, both 45, with children aged 15 and 12 and $114,000 of wages in 2019. Against HHS's guideline for four, $28,870, the family's income comes to 394% of the poverty line, under the credit's 2019 limit of 400%. The model divides by $28,390, places the family at 401% and gives no credit. When we set Hawaii's 2018 additional_person to 4,970 in a reform, we estimate a credit of $9,519 for the same family.

Where the model sets it

The model sets Hawaii's 2018 additional_person to 4_810 on line 97 of fpg.yaml, the same value as 2017 on line 96. fpg() in tax_unit_fpg.py and spm_unit_fpg multiply it by the number of people after the first. aca_magi_fraction divides MAGI by the prior year's guideline.

Reproduction

With policyengine-us 2.117.4 (main at 935b24f):

from policyengine_us import Simulation

years = ["2018", "2019"]


def hawaii_family(wages):
    return {
        "people": {
            "a": {"age": {y: 45 for y in years}, "employment_income": {"2019": wages}},
            "b": {"age": {y: 45 for y in years}},
            "c": {"age": {y: 15 for y in years}, "is_tax_unit_dependent": {y: True for y in years}},
            "d": {"age": {y: 12 for y in years}, "is_tax_unit_dependent": {y: True for y in years}},
        },
        "tax_units": {"t": {"members": ["a", "b", "c", "d"]}},
        "spm_units": {"s": {"members": ["a", "b", "c", "d"]}},
        "families": {"f": {"members": ["a", "b", "c", "d"]}},
        "households": {"h": {"members": ["a", "b", "c", "d"], "state_code": {y: "HI" for y in years}}},
    }


for wages in [110_000, 114_000]:
    sim = Simulation(situation=hawaii_family(wages))
    print(
        wages,
        sim.calculate("tax_unit_fpg", 2018)[0],
        sim.calculate("aca_magi_fraction", 2019)[0],
        sim.calculate("aca_ptc", 2019)[0],
    )

110000 28390.0 3.87 9912.914
114000 28390.0 4.01 0.0

We set the household's state to Hawaii for both 2018 and 2019. If a user gives the state for 2019 alone, aca_magi_fraction reads the default state for 2018 and applies the contiguous states' guideline, a separate defect (#9482, with a fix open in #10860).

Possible fix

Set Hawaii's 2018 additional_person on line 97 to 4_970, and add a tax_unit_fpg test that expects $18,930 for a Hawaii unit of two in 2018.